The long-term objective of this application is to develop methods for the efficient determination of atomic-resolution three-dimensional (3D) structures of large biological complexes in their native, non-crystalline states by electron cryomicroscopy (cryoEM). The emerging technology of cryoEM and 3D reconstruction offer great promise for structural studies of supramolecular machines that are difficult to study by X-ray crystallography or NMR. The PI's group has published the cryoEM-determined structures of a number of complexes at subnanometer resolutions, including the 6-Angstrom structure of rice dwarf virus (RDV), which was subsequently confirmed by X-ray crystallography. Because atomic-resolution cryoEM images have been recorded using state-of-the-art instruments, we hypothesize that powerful computation and data mining tools can be developed to process terabytes of noisy image data for determining atomic models of large complexes by cryoEM. thus significantly enhancing the value of cryoEM structures. The overall goal of this exploratory project is to build upon our initial success of subnanometer cryoEM studies to develop efficient and accurate methods for determining near-atomic resolution 3D maps from cryoEM images and for building atomic models from such maps. First, several novel computational methods will be developed to improve the accuracy and efficiency of orientation and center estimation and refinement, and to allow full contrast transfer function correction associated with the inherent depth-of-focus problem of large complexes. Second, data management solutions, structure mining and atomic model-building tools will be implemented and integrated with other disparate bioinformatics tools under a user-friendly interface of the IMIRS package to tackle the inevitable and daunting tasks associated with high-resolution cryoEM reconstructions. To eliminate potential bias inherent in method developments using simulated data, our new methods will be subjected to rigorous and unbiased testing and validation by determining the atomic structures of RDV, an ideal model system substantially studied by the Principal Investigator. This project will result in a full spectrum of efficient and effective algorithms and software tools that will be useful and freely available to the broad areas of structural and computational studies of other supramolecular assemblies. Our study fits well in two of the three themes of the NIH Roadmap initiatives: research in structural, bioinformatics and computational biology under the theme of New Pathways to Discovery and interdisciplinary research under the theme of Research Teams of the Future.